Effect of palladium addition on nickel silicide ... - SPIE Digital Library

Effect of palladium addition on nickel silicide formation on Si (111) Aliaksandra O. Karabko1, Anatoly P. Dostanko1, Sergei M. Zavadsky1, Jinfang Kong2, Wenzhong Shen2 1Department of Electronic Technique and Technology, Belarusian State University of Informatics and Radioelectronics, 220013 Minsk, Belarus, [email protected] 2Department of Physics, Shanghai Jiao Tong University, 200030 Shanghai, China ABSTRACT In this work, the influence of palladium addition on phase transition, surface morphology, structural, vibrational, and electrical properties of nickel silicide is investigated at various temperatures. For Ni(Pd)Si films micro-Raman measurements have yielded Raman phonon peaks belonging to NiSi phase, although redshifted, on par with new peaks at 322 and 434 cm-1, not determined before, which we assign to the compositional disorder, introduced by Pd. The results have shown that Ni(Pd)Si films are thermally stable up to 900 °C, which is 100-150 °C more than that for pure NiSi films. Applying Miedema’s model we have calculated the heat of formation for Ni(Pd)Si and found it to be more negative than that for pure NiSi, revealing a key role of Pd in the retardation of NiSi2 phase formation. AFM results have shown that the presence of Pd favorably influences the surface morphology of NiSi, resulting in a smoother surface. Furthermore, we have discussed the impact of annealing conditions on peculiarities of Pd diffusion, element distribution and electrical properties of Ni(Pd)Si and NiSi films. Keywords: NiSi, Ni(Pd)Si, Raman spectroscopy, AES, AFM, thermal stability

1. INTRODUCTION Recently, the properties of NiSi have been studied extensively1. The use of NiSi in microelectronic devices has numerous advantages such as low formation temperature and compatibility with silicon and germanium, low silicon consumption during silicidation, low contact resistance. However, there are two main degradation mechanisms for NiSi films annealed at high temperatures: the agglomeration of the film and the phase transition to NiSi2 phase, which is highly resistive. Moreover, because of this degradation a very rough interface would be resulted between the silicide layer and the silicon substrate and such roughening leads to anomalously large junction leakage current. One of the main modifications done to nickel silicide to improve its properties is the introduction of alloying elements that can substantially affect the morphological and formation stability of NiSi compound2, and, therefore, they are very effective in the enhancement of the temperature range in the postsilicide treatment. In the past five years, the modification of nickel silicide composition was performed by adding Pt, Zr, W, Mo, and Pd2, with Pt being the most studied additive. Hence, the further thorough investigation of the other elements’ influence on the properties of Ni-Si system, for example such as Pd, is appealing and needs to be carried out. Nevertheless, one has to be aware of the possible effects on silicide’s structure and properties, which an introduction of the alloy element can induce. Furthermore, the question arises on the fast and adequate methods of the evaluation of peculiarities and modifications done in the modified NiSi. A number of experimental techniques are known to provide the accurate information about the structure, element distribution, and phase formation in NiSi. Such can be expensive and complex XRD, RBS and TEM techniques. We have chosen techniques that are less time consuming and do not require any sample preparation, which were micro-Raman spectroscopy, Auger electron spectroscopy (AES) and atomic force microscopy (AFM).

2. EXPERIMENTAL AND MATERIALS ANALYSIS DETAILS Thin Ni films with the addition of 10 at. % of Pd and for comparison pure Ni films were deposited on n-type (111) Si by dc magnetron sputtering system. Magnetron sputtering was performed from a nickel source of 99.99 % purity and from a compositional target made of a nickel source of 99.99 % purity and of a palladium source of 99.95 % purity. The thickness of the deposited film, comprising either Ni or the mixture of Ni and Pd, was about 100 nm for both kinds of samples. The native oxide was removed by ion cleaning treatment of the surface before the film sputtering process began. The pure nickel silicide and nickel with palladium silicide films were formed by vacuum annealing of Ni and Ni(10 at. % Pd) films on silicon substrates in inert gas ambient. We have carried out the two-step annealing process with Twelfth International Workshop on Nanodesign Technology and Computer Simulations, edited by Alexander I. Melker, Vladislav V. Nelayev, Proc. of SPIE Vol. 7377, 73770W © 2009 SPIE · CCC code: 0277-786X/09/$18 · doi: 10.1117/12.836983 Proc. of SPIE Vol. 7377 73770W-1

the first annealing at 475 °C for 30 min to form nickel silicide with a subsequent etching of the unreacted metal. Annealing temperatures for the second annealing were varied up to 900 °C and also for 30 min. 0

Raman Intensity (arb. units)

NiPd/Si

214 291 362 196

900 C 0

NiSi2

Ni/Si 210 193

292

200

0

600 C

400

362

600 0

900 C

NiSi2 322

850 C 0 800 C 0 750 C

0

434

850 C 0

800 C

0

600 C

200

400

-1

Raman Shift (cm )

600

Fig. 1. Room-temperature micro-Raman spectra for Ni(Pd)Si and NiSi (inset), annealed at various temperatures The phase transformation and vibrational properties of the pure and alloy nickel silicides were studied by the microRaman technique. Raman measurements were carried out in a backscattering geometry of z(x,-) z on a Jobin Yvon LabRam HR 800 UV micro-Raman system with Andor DU420 CCD detector and a Linkam THMS600 temperature stage using a 514.5 nm line of Ar+ laser as an excitation source. The temperature range for the temperature-dependent Raman was chosen from 83 to 563 K with a step of 30 K. The composition of the nickel silicide thin films in both cases was studied using Auger electron spectroscopy (AES) depth profiles. AES measurements were conducted on Perkin Elmer spectrometer PHI-660 at room temperature using argon ion sputtering. The surface roughness and grain sizes of nickel silicide films were studied using atomic force microscopy (AFM). AFM scans were obtained using a NT-206 scanning probe microscope. The measurement of the sheet resistance of the pure and alloy nickel silicides was carried out by means of a four-point probe technique using the device IUS-3M. In Fig. 1 the results of room-temperature micro-Raman measurements for the Ni(Pd) film on Si substrates, annealed at various temperatures are displayed. To get the accurate details of Raman study of Ni(Pd)Si and NiSi systems please refer to our recent work3. NiSi belongs to the orthorhombic structure (MnP-type, space group Pnma, D2h16) and is Raman active. For annealing temperatures from 475 up to 850 °C we observe NiSi phase with the distinctive Raman peaks at 193, 210, 292, 362 cm-1. It is noted, that the most intensive peaks at 193 and 210 cm-1 are redshifted in comparison with the peaks at 196 and 214 cm-1 for pure NiSi films (please see the inset in Fig. 1), which is due to the change of the interatomic distance, caused by the presence of Pd. Starting from the annealing temperature of 800 °C we also detect peaks at 322 and 434 cm-1, unregistered in the literature before. We suggest that the new phonon peaks at 322 and 434 cm-1 can be attributed to the disorder Raman effects, induced by the addition of Pd element. In fact, the presence of the alloy introduces the defects to the films, which can result in the breakage of the translational symmetry, and as a result, the appearance of more peaks in Raman spectra. From Fig. 1 it is obvious that a full transformation to NiSi2 phase for Ni(Pd)Si sample occurs at 900 °C as we observe diffuse, broad peaks in the range of 200-450 cm-1, which are characteristic to NiSi2 phase4. The experimental results obtained in our investigation show that the introduction of only 10 at. % of Pd to the initial nickel film content can retard the formation of NiSi2 phase up to 900 °С, which is generally about 100-150 °С more than that for pure NiSi films.

Proc. of SPIE Vol. 7377 73770W-2

3. MICRO-RAMAN CHARACTERIZATION OF Ni(Pd)Si

(b)

563K 533K 503K 473K 443K 413K 383K

Raman Intensity (arb. units)

(a)

353K 323K

o

NiSi, 600 C

293K 263K 233K 203K 173K 143K 113K 83K o

NiPdSi, 600 C

200 225 250 200 225 250 -1

Raman Shift (cm ) Fig. 2. Temperature-dependent Raman spectra for (a) NiSi and (b) Ni(Pd)Si Further, we focus on the investigation of temperature dependencies of frequencies and damping of the intensities for the strong Raman peaks of Ni(Pd)Si (ω = 210 cm-1) and NiSi (ω = 214 cm-1), annealed at 600 °C, to understand the contribution of Pd addition, the details of which are presented in Ref. 3. Fig. 2 shows Raman spectra within the measured temperature in the range of 83-563 K. In order to accurately determine the Raman frequency of the modes under investigation we apply Lorentz model to fit the Raman peaks throughout the measured temperatures. With the increase of the temperature, Raman spectra clearly indicate a non-linear decrease in phonon peak position of about 10 cm-1 (from 217.8 to 208.0 cm-1) for NiSi, displayed in Fig. 2(a), and of about 6 cm-1 (from 215.0 to 209.4 cm-1) for Ni(Pd)Si, which is seen in Fig. 2(b), both samples initially annealed at 600 °C. The reduction of ~ 4 cm-1 in the Raman frequency downshift demonstrates the favorable use of Ni(Pd)Si for microelectronics applications. We can quantitatively analyze the experimental peak redshift for Ni(Pd)Si and NiSi. The downshift of the Raman frequency with the increase of temperature can be explained by the combination of the effects of anharmonic coupling to other phonons, thermal expansion, and lattice-mismatch-induced strain between thin films and substrates. In modeling the Raman shift, the temperature-dependent Raman frequency ω(T) due to various factors can be written as5:

ω (T ) = ω 0 + Δω d (T ) + Δω e (T ) + Δω s (T )

(1)

with ω0 the perfect harmonic lattice phonon frequency of the optical mode, Δωd(T) the contribution of the anharmonic coupling to phonons of other branches, Δωe(T) the one from thermal expansion of the lattice or volume change, and Δωs(T) the thermal mismatch between the silicide film and the silicon substrate. The term Δωd(T) can be described by:

⎡ ⎤ ⎡ ⎤ ⎢ ⎥ ⎢ ⎥ 2 3 3 Δωd (T ) = A ⎢1 + + B ⎢1 + + ⎥ ⎥, hω hω hω ⎢ exp( 0 ) − 1 ⎥ ⎢ exp( 0 ) − 1 [exp( 0 ) − 1]2 ⎥ 2kT 3kT 3kT ⎣ ⎦ ⎣ ⎦


(2)

where the first term is the decay of zero-center phonons into two phonons (three phonon coupling), the second one represents the decay into three phonons (four-phonon coupling). Anharmonic constants A and B are selected as fitting parameters. We have considered the symmetric decays of zone-centre phonons into two and three phonons with frequencies ω0/2 and ω0/3, respectively. The term Δωe(T) can be given as: T ⎧⎪ ⎡ ⎤ ⎫⎪ Δωe (T ) = ω0 ⎨exp ⎢ −γ g ∫ [α a (T% ) + α b (T% ) + α c (T% )]dT% ⎥ − 1⎬ , ⎪⎩ 0 ⎣ ⎦ ⎪⎭

(3)

~

where αi ( T ), i = a, b, c - thermal coefficients of linear and γg — Grüneisen constant. There is a noticeable lattice and thermal mismatches between Ni, Ni(Pd)Si and Si, which could result in the straininduced contribution Δωs(T) to the phonon frequency shift. The different thermal expansion coefficients of a thin film and a substrate lead to a temperature-dependent in-plane strain ε(T)5: T

ε (T ) = (1 + ε g )

1 + ∫ α a,s (T% )dT% Ts

T

−1

,

(4)

1 + ∫ [α i (T% ) + α j (T% )]dT% Ts

where αa,s and αi, αj (i=a, b, c; j=b, c, a) are the temperature-dependent in-plane linear expansion coefficients of the substrate and film, respectively, and εg the residual strain in the silicide film at silicidation temperature Ts, which is assumed to be zero6. Since the phonon deformation potentials and elastic constants data of nickel monosilicide films are not available up to now, we set them as one unified constant C, which is also a fitting parameter in our calculations. Therefore, the strain-induced contribution Δωs(T) to the Raman shift of the frequency is:

ΔωS (T ) = Cε (T ) .

(5)

After modeling the experimental data using Origin 7.0, we note that the present model describes well the downshift of the phonon frequency, yielding the reasonable value of C = 7.0 cm-1, together with the results of A, B, and ω0 listed in Table 1. The ratio A/B allows us to assess the type of the decay process and we conclude the mechanism of the zonecentre optical phonon decay into the lower energy phonons to be mainly a three-phonon process for pure NiSi, as A/B equals to 115.0. For Ni(Pd)Si annealed at 600 °C the ratio of A/B is much smaller (≈1.6), indicating the possibility of both three-phonon and four-phonon decay processes. It is due to the incomplete formation of solid solution Ni1-xPdxSi at 600 °C, and the consequent compositional disorder in the films, that results in the possibility of both mechanisms of phonon decay. Compound NiSi Ni(Pd)Si

A, cm-1 -1.043 -0.100

Table 1 - Fitted values of A, B and ω0 according to modeling of the experimental B, cm-1 A/B ω0, cm-1 -0.009 115 219.5 -0.063 1.6 215.3

Evaluating the independent contributions of the terms Δωd(T), Δωe(T), and Δωs(T) from equation (1) to the Raman frequency downshift and phonon softening observed, we found that the lattice and thermal mismatch-induced strain is small and seems not to affect the Raman frequency downshift for all samples and we conclude that the phonon decay Δωd(T) has the major impact on frequency downshift for NiSi and Ni(Pd)Si films with the minor role of thermal expansion Δωe(T). It was previously assumed6 that the phonon decay process is a dominant process to contribute to the Raman frequency downshift in pure NiSi films (the impact of the thermal expansion and strain-induced lattice-mismatch has not been considered at that time).


4. CALCULATION OF THE HEAT OF FORMATION OF Ni(Pd)Si The enhancement of the thermal stability for the alloy nickel silicide such as Ni(Pd)Si can be explained by the comparison of the values of heats of formation of nickel monosilicide phase in Ni(Pd)Si and NiSi films. We have applied the semi empirical Miedema’s model7 to calculate the heat of formation of nickel monosilicide in Ni(Pd)Si film and to compare with that of the pure NiSi sample, although the heat of formation for NiSi films could be obtained directly. For Ni(Pd)Si films a direct assessment is very difficult as we lack adequate thermodynamics data. The semi empirical model of Miedema for metal cohesion is based on two parameters φ* and nws1/3 that are assigned to each atom8, 9. In this model it is also assumed that there are two contributions to the heat of formation for a binary alloy system. One contribution is attractive and arises from charge transfer between neighbouring cells and is proportional to - (Δφ*)2, the other contribution is repulsive and arises from a surface-tension term and is proportional to (Δn1/3)2. The detailed description of the Miedema’s model is presented elsewhere7, 8. The resulting equation for evaluating the heat of formation of nickel monosilicide phase is: ΔH = c A f BA

V A2 / 3

−1 / 3 ) (nwS av A S S S f B = cB [1 + 8(c AcB )2 ]

c SA =

c AV A2 / 3

P[−(Δφ * ) 2 +

c AV A2 / 3 + cBVB2 / 3

Q 1 / 3 ) 2 − R / P] , (ΔnWS P

(6) (7)

, c BS =

c B V B2 / 3 c AV A2 / 3 + c B V B2 / 3

(8)

where ΔH - the heat of formation, cA, cB - the concentration of (A) Si and (B) Ni, respectively, cAS, cBS - surface concentrations of Si and Ni, VA, VB - the molar volume of Si and Ni, nws1/3 - the density of electrons at the boundary of Wigner-Seitz cells of pure atoms, φ*-the chemical potential for pure elements, fBA - the multiplicative factor for ordered alloys, P, Q, and R - constants. Here φ* is expressed in V, V in cm3, nws in density units (d.u., 1 d.u. = 6×1022 el/cm3), ΔH in kJ/mol, P, Q/P, R/P are equal to 12.45 (the unit is found in Ref. 7), 9.4 V2/(d.u.)2/3 and 2.1 V2, respectively. The results of the calculations yield the following values if the heat of formation that are presented in Table 2. We have performed the calculations for the addition of about 10 at. % of Pd to nickel film content. Moreover we have carried out the same kind of estimation for the addition of 10 at. % of Pt to nickel film content, the positive effect of the introduction of which is already known8. Compound NiSi Ni(Pd)Si Ni(Pt)Si

Table 2 - Calculated heat of formation for the nickel monosilicide phase in pure and alloyed nickel silicides ΔH, kJ/mol -92.65 -96.68 -95.57

The calculated heat of formation is found to be 4 kJ/mol more negative for Ni(Pd)Si than that for NiSi sample, which is the reason for the retardation of NiSi2 formation we observe in Raman spectra. Furthemore, it is necessary to note, that the heat of formation of nickel monosilicide phase in Ni(Pd)Si film is more negative than that for Ni(Pt)Si, which indicates that Pd introduction is the most effective for the enhancement of the thermal stability of the nickel monosilicide film.

5. COMPOSITION ANALYSIS We have also performed a thorough analysis of element distribution profiles in Ni(Pd)Si and NiSi films to follow the silicidation process at various temperatures by means of Auger electron spectroscopy. The composition ratios of the elements during different steps of the silicidation are displayed in Fig. 3. At annealing temperature of 475 °C for NiSi film, the composition of elements is uniform, and elements are homogeneously distributed throughout the formed silicide layer, while for Ni(Pd)Si film one can note at least two different composition ratios of elements in the silicide layer. The maximum of Pd distribution in the film is observed at about 0.1 µm for this sample. This can be explained by the fact that the initial thickness of the deposited metal alloy layer was also about 0.1 µm, which means that at this depth the


interface between silicon and the metal film has occurred. By solid phase formation process such interface is often the main source of defects and impurity localization during the beginning stages of annealing. At 750 °C, Auger electron spectroscopy results yield a composition of elements to be straight 50 % Ni to 50 % Si throughout the formed silicide layer in NiSi sample. At the same time according to Raman measurements (Fig. 1, inset) this temperature is regarded a critical transition temperature, when both phases NiSi and NiSi2 are observed, although NiSi2 concentration is still small. At higher annealing temperatures we observe the slight dominance of silicon over nickel in the spectra, yielding the composition of about 60 % Si to 40 % Ni, respectively, while only NiSi2 phase is detected by Raman spectroscopy. As Raman measurements unambiguously identify the phase content, we can conclude that such discrepancies in Raman and Auger electron spectroscopy results can be due to the deviation from the equilibrium stoichiometry in the compound to the slight lack of silicon and the excess of metal phase. The latter is undetectable by Raman measurements due to the fact that the mobile charge carriers in metals are electrons that absorb the laser power and do not give rise to any peaks in Raman spectra. 100

90 80 70 60 50 40 30 20

100

a Ni Si 0.4

0.3

0.2

0.1

Ph

100

901

RhI

0.3

0.5

0.4

Si

0.3

0.4

0.5

0.6

0.7

.

T=700 °C

801 70 60

Ni

Si

Ni

Si

Ni 0.2

0.3

0.4

0.5

0.6

T=750 °C

d

100 90 80 70 60 50 40 "U 20 10 0

Si

Si

Ni

10

0

0.3

0.2

0.1

0.4

0.2

0.1

0.

Ni

40 30 20

0.3

0.

t

Ni

Pd 0.4

0.3

0.2

0.1

0.5

0.5

100 90 80 70 60 50 40 30 20

In i

i

I

0.1

100 90 80 70 60 50 40 30 20 10 0

0.2

0.1

100 90

C T=700°C

060 50

100 90 80 70 60 50 40 30 20 It)? 0

0.5

'

'i=obc'

0

0.1

70

.' '2-1 0.4

10

0.2

100

o90

1

Ni 0.3

0.2

70 60 50 40 30 20

0.1

©N 80

Si

0.1

b T=600°C

Si

Ni

0.5

70 60 50 40 30 20 10 0 100 90 80 70 60 50 40 30 20 10 0

T=475 °C

Pd 1------

10 0

100 90

.

80 70 60 50 40 30 20

-

.

0.2

I

I

0.3

I

0.4

-----0.5

I;

0.2

0.1

0.6

0.3

0.5

0.4

100

f

90 80 70 60 50 40 30 20

Si

Ni

m T=850°C

100 90 80 70 60 50 40 30 20

0.2

0.3

0.4

0.5

100 90 80 70 60 50 40 30 20

Si

Ni

Ni

0.2

0.4

0.6

0.8

1.0

0.5

Si

Si

Ni

Ni

Pd

0

0

0.4

T=900°C

10

10

0.3

0.2

0.1

0.6

T=900 °C

Ni

Pd

10 0 0.1

Si

0.1

0.2

0.3

Depth, Rm

0.4

0.5

0.6

0.7

Fig. 3. AES depth profile of NiSi (a, b, c, d, e, f, g) and Ni(Pd)Si (h, j, i, k, l, m, n) samples after annealing at various temperatures


Concentration of Pd, at. %

The same conclusion can be reached also for Ni(Pd)Si films, where at the temperature of 750 °C the element composition is 46 % Si, 4 % Pd, and 50 % Ni, which also indicates the dominance of the metal phase. Therefore, we can say that there is a compositional disorder in both Ni(Pd)Si and NiSi samples to the side of the excess of metal phase. However, at the highest annealing temperature the depth profile for Ni(Pd)Si indicates that there is a uniform layer of the almost accurate NiSi2 composition of 65 % Si to 31 % Ni. The distribution of Pd in NiSi layer can be described by Gauss function at lower annealing temperatures up to 700 °C. At higher temperatures Pd becomes uniformly distributed through out the silicide layer. Furthermore the initial concentration of 10 at. % of Pd in 100 nm nickel film is reduced to about 4 at. % of equally distributed Pd in silicide film, changing by exponential decay law with temperature (Fig. 4). Considering that 1 nm of metal results in 2.34 nm of nickel monosilicide silicide such distribution of Pd seems to be reasonable. 10 9 8 7 6 5 4 400

500

600

700

800

900

0

Temperature, C

Fig. 4. Temperature dependence of the concentration of Pd in nickel silicide layer Therefore, by comparison of phase transformation kinetics and the change in the element distribution it was noted that either both annealing conditions and the presence of Pd result in extra sedimentation of a small quantity of nickel metallic phase, which that the formed silicides are not fully stoichiometric ideal.

6. ELECTRICAL PROPERTIES OF NiPdSi AND NiSi One of the quick, but a rather rough estimation of phase transition and electrical properties of silicides can be done by measuring the sheet resistance. Estimating the value of the sheet resistance is an express-control method in silicidation process10 and is regarded to be the criteria of the electronic quality of the sample and its correspondence to NiSi or NiSi2 phase. A continuous nickel silicide film has a constant value of the sheet resistance, while the film with the agglomeration zones has a much higher one. The values of the sheet resistance of the formed NiSi and Ni(Pd)Si films are measured using the four-point probe technique and are displayed in Table 3. Тable 3. The sheet resistance of NiSi and Ni(Pd)Si

ρs, Ω/

T, °C 475 600 700 750 800 850 900

NiSi 0.50 0.53 0.66 0.78 2.60 2.61 2.63


Ni(Pd)Si 1.15 1.10 1.16 1.13 1.10 1.44 2.77

a)

b)

c)

d)

e)

f)


g) h) Fig. 5. Comparative AFM images of NiSi (a-d) and Ni(Pd)Si (e-i) surfaces at various annealing temperatures: а, e - Т = 475 °С; b, f - Т = 600 °С; c, g - Т = 750 °С; d, i - Т = 850 °С The analysis of the results shows that NiSi films have low values of the sheet resistance in the temperature range of 475750 °С. According to the measured values of the film thickness the resistivity was found to be 13.5-18 μΩcm, which corresponds well to the values obtained in literature11. In the same range of temperatures the Ni(Pd)Si films have the higher values of the sheet resistance, as it is seen from Table 3. The higher values of the sheet resistance could be due to the presence of an additional metal (Pd), resulting in an initially higher value of the resistance. It is explained by the fact that at interaction of two types of conductors (metals) a series connection is formed, and therefore the values of the resistance are summed up. As it is seen from Table 3, a sharp increase (in 2-3 times) of the sheet resistance is noticed in the result of annealing of the NiSi film at temperatures higher than 750 °С, which is due to the transformation of the monosilicide phase to the disilicide phase. For Ni(Pd)Si sample we can conclude that nickel monosilicide phase is stable up to 850 °С, as the sheet resistance does not alter much. At 900 °C it increases in two times, indicating that the phase transformation process has started or a severe agglomeration occurred. Hence, in the extended temperature range it is noticeably better to use Ni(Pd)Si film as its resistance is more stable at higher temperatures.

7. SURFACE MORPHOLOGY ANALYSIS OF NiSi AND NiPdSi With the aim of studying the change of the surface morphology as the result of annealing we have investigated the samples of NiSi and Ni(Pd)Si films by means of atomic force microscopy. AFM scan results are presented in Fig. 5. Considering the change of the surface topology we can judge over the change of the phase composition of the silicides and the surface homogeneity of the samples. As it is seen from Fig. 5 the order of roughness of the surface for Ni(Pd)Si samples (Fig. 5e-i) is lower in two or more times than that of NiSi, indicating that the addition of Pd in the composition is favorable. In the temperature range of annealing, when the silicide phase is a monosilicide, the surface is composed of separately positioned small peaks. Moreover, the peaks for Ni(Pd)Si films are densely situated to each other and are smaller in size. It is necessary to note, that during the transformation of nickel monosilicide to nickel disilicide the surface morphology becomes more rough and peaks are changed into columns (Fig 5 c, i), which correlates with the data previously appeared12.

8. CONCLUSIONS In this study, we have performed a thorough investigation of the effect of palladium addition to nickel silicide. The results reveal that the thermal stability of nickel monosilicide phase can be substantially enhanced up to 900 °C by the introduction of Pd into silicide content, which we have proved by micro-Raman and the sheet resistance measurements. The calculations performed for Ni(Pd)Si according to semi empirical Miedema’s model have shown that the heat of formation for this system is about 4 kJ/mol less than that for NiSi, giving the reason and additionally proving the possibility of the thermal stability enhancement of NiSi monosilicide phase. Some peculiarities in Raman vibrational spectra for Ni(Pd)Si were noticed: the introduction of Pd increases the interatomic distance, which in turn shifts the vibrational peak positions to lower frequencies in comparison with that in pure NiSi. At the same time the appearance of


new peaks in Raman spectra are detected at temperatures higher than 700 °C, which we assume is due to the compositional disorder effects in Ni(Pd)Si. Also we have demonstrated that the temperature-dependent Raman study of phonon dynamics and calculations performed for both Ni(Pd)Si and NiSi prove that the main mechanism for the observed total frequency peak downshift is due to the anharmonic phonon decay into two phonons with a minor influence of the thermal expansion of the lattice. The further notice of great importance has been done on the peculiarities of Pd distribution in the silicide layer, where a uniform distribution of Pd in nickel silicide layer is reached only at 700 °C. Moreover, we found the deviation from exact silicide stoichiometry to the presence of the excess nickel metallic phase, which needs to be further investigated. The AFM scan results have shown that Ni(Pd)Si films have a smoother surface, with a value of a square mean root roughness several times lower than that of NiSi films.

REFERENCES 1. C. Lavoie, F.M. d’Heurle, C. Detavernier and C. Cabral. Jr., “Towards implementation of a nickel silicide process for CMOS technologies”, Microelectronic Engineering, 70, 144 - 157 (2003). 2. C. Lavoie, C. Detavernier, C. Cabral Jr., F.M. d’Heurle and A.J. Kellock, “Effects of additive elements on the phase formation and morphological stability of nickel monosilicide films”, Microelectron. Eng., 83, 2042 - 2054 (2006). 3. A.O. Karabko, A.P. Dostanko, J.F. Kong and W.Z. Shen, “Phase Transition and Phonon Dynamics in NiPdSi: An Annealing Study by Temperature-Dependent Raman Spectroscopy”, J. Appl. Phys., 105, 033518 (2009). 4. F. Li, N. Lustig, P. Klosowski and J.S. Lannin, “Disorder-induced Raman scattering in NiSi2”, Phys. Rev. B, 41, 10210 (1990). 5. X.D. Pu, J. Chen, W.Z. Shen, H. Ogawa and Q.X. Guo, “Temperature-dependence of Raman scattering in hexagonal indium nitride film”, J. Appl. Phys., 98, 033527 (2005). 6. S.K. Donthu, S. Tripathy, D.Z. Chi and S.J. Chua, “Raman scattering probe of anharmonic effects in NiSi”, J. Raman Spectrosc., 35, 536 - 540 (2004). 7. A.R. Miedema, P.F. deChatel, and F.R. de Boer, “Cohesion in alloys - Fundamentals of a semi-empirical model”, Physica B, 100, 1 - 28 (1980). 8. R.N. Wang, Y. He and J.Y. Feng, “Explanation of the enhancement of NiSi thermal stability according to TFD equations and Miedema’s model”, NIM B, 222, 462 - 468 (2004). 9. J.R. Chelikowsky, “Microscopic basis of Miedema’s theory of alloy formation”, Phys. Rev. B, 25, 6506 - 6508 (1982). 10. S.P. Murarka, Silicides for VLSI, Academic press, New York, 1986. - 176 p. 11. H. Iwai, T. Ohguro and S-I. Ohmi, NiSi salicide technology for scaled CMOS, Microelectron. Eng., 60, 157 - 169 (2002). 12. S. Abhaya, G. Amarendra, S. Kalavathi, P. Gopalan, M. Kamruddin, A.K. Tuagi, V.S. Sastry and C.S. Sundar, “Silicidation in Ni/Si thin film system investigated by x-ray diffraction and Auger electron spectroscopy”, Applied Surface Science, 253, 3799 - 3802 (2007).
